Forecasting the Chinese stock volatility across global stock markets

Abstract

In view of the growing concern of Chinese stock fluctuation, this paper forecasts the Chinese stock volatility by extracting global stock information by combining forecasts of time-varying parameter (TVP) volatility models. First, we construct individual constant coefficient (CC) models and TVP models across 27 global stock markets, and then use several strategies to combine their forecasts. The results show global stock information does forecast the future volatility of Chinese stock market. Both the forecast accuracy and economic values can be further improved by using strategies combining TVP models with global stock information. Specifically, the median combination shows its superiority for the volatile Chinese stock market. Our findings are robust to different estimation window sizes, volatility proxies, and evaluation criteria.